UCLA UCLA Previously Published Works

Transcription

1 UCLA UCLA Previously Published Works Title Food requirements of wild animals: predictive equations for free-living mammals, reptiles, and birds. Permalink Author Nagy, K Publication Date Peer reviewed escholarship.org Powered by the California Digital Library University of California

2 1 Reference citation: Nagy KA (2001) Food requirements of wild animals: predictive equations for free-living mammals, reptiles, and birds. Nutrition Abstracts and Reviews, Series B 71, 21R-31R. Food requirements of wild animals: Predictive equations for free-living mammals, reptiles, and birds Kenneth A. Nagy Department of Organismic Biology, Ecology and Evolution, University of California, Los Angeles, California, USA, ; kennagy@biology.ucla.edu Key words allometry, bioenergetics, diet effect, dry matter intake, feeding rate, fresh matter intake, habitat effect, scaling Abstract Feeding rates (intake of both dry matter and fresh matter) by 79 species of mammals, 95 species of birds and 55 species of reptiles were estimated from doubly labeled waterbased measurements of field metabolic rate on each species (Table 1). Allometric (scaling) regression analyses of log10-transformed feeding rates vs. body mass yielded statistically significant relationships for 90 different taxonomic, dietary and habitat groupings of species. The resulting exponential equations can be used to predict the daily food requirements needed to maintain energy balance for free-living mammals (Table 2), birds (Table 3), and reptiles (Table 4) with an average error of about 5% to 60%, depending on the group. The ability to predict feeding rates of terrestrial vertebrates should be useful to zoo keepers, animal nutritionists, veterinarians, pet hobbyists, wildlife zoologists, game managers, range biologists, preserve directors and planners, conservationists, paleontologists and ecosystem modelers. These equations should

3 2 underestimate somewhat the feeding rates of free-living animals that are growing, reproducing or storing up fat. The equations probably overestimate the feeding rates of captive wild animals (e.g. in zoos) and of free-ranging animals during some phases of their lives when they either do not or cannot feed normally. Introduction One of the first questions people ask about a wild animal is What does it eat? Those who work with animals also want to know How much does it eat each day? Zoo keepers, animal nutritionists, veterinarians, pet hobbyists, wildlife zoologists, game managers, range biologists, preserve managers, conservationists, paleontologists and ecosystem modelers are among the people that are concerned about the daily food needs of different species of living and extinct mammals, birds and reptiles. The utility of such information ranges from practical applications to theoretical evaluation of the role of vertebrate consumers in models of the biosphere. Early estimates of the food needs of wild animals were based on laboratory measurements of rates of oxygen consumption or carbon dioxide production (indirect calorimetry). Corrections to account for the differences between metabolic rates measured in captivity and those in the field were problematic, and largely conjectural. Fortunately, the advent of the doubly labeled water method (Lifson and McClintock,1966) has made it possible to measure carbon dioxide production in freeliving, air-breathing vertebrates in their natural habitats. The field metabolic rates (FMRs) of over 229 species of terrestrial vertebrates have now been determined with this technique. The size of the animal, expressed as body mass, explains most of the difference in whole-animal FMR between species, with larger animals generally (but not always) using more total energy each day than do smaller ones. Taxonomic Class (mammal, bird, reptile) explains much of the remaining difference between species. Thus, allometric or scaling analyses (log10 FMR in kilojoules metabolized per day versus log10 body mass in grams) indicate that, within each Class, log body mass explains about 94% of the variation in log FMR between species (Nagy et al., 1999). The equations

4 3 describing these allometric relationships can be used to predict the FMRs of species that have not yet been studied. The food requirement of an animal can be estimated from its energy requirement by calculating the amount of food needed to provide that amount of metabolisable energy. This review includes allometric equations for predicting both dry matter and fresh matter intake rates for wild reptiles, birds, and mammals living in their natural habitats, as derived from FMR measurements along with information about dietary energy content. Animals that are held captive, as in zoos, corrals or cages, will probably but not necessarily have lower daily food needs than those estimated from the equations herein, due to lower activity levels and more benign microclimates than those they experience in nature. Methods Field feeding rates were estimated from field metabolic rates, as measured using the doubly labeled water method (Lifson and McClintock, 1966; Nagy, 1983; Speakman, 1997) for the 229 species of terrestrial vertebrates summarized in the recent review by Nagy et al. (1999; see link for references to individual studies). The FMR for a species, in units of kj/d, was divided by the metabolisable energy content of its diet, either in units of metabolisable kj/g dry matter or in units of metabolisable kj/g fresh matter, to calculate feeding rates in units of g dry matter intake (DMI)/d or in units of g fresh matter intake (FMI)/d. Metabolisable energy, as used in this review, is defined as gross food energy minus energy excreted as feces and urine, and values based on dry matter for the various diets were taken from Nagy et al. (1999). These values were converted to units of fresh matter using average dietary water content values of 66% for insects, 70% for a carnivore s diet, 67% for green plant matter, 68% for an omnivore s diet, 10% for dry seeds, 76% for nectar, 73% for fruit, and 73% for fish (from Nagy and Peterson, 1988).

5 4 The conversion factors used were: mammalian insectivore (having urea excretion), 18.7 kj/g DMI and 6.17 kj/g FMI; bird and reptile insectivore (having uric acid excretion), 18.0 kj/g DMI and 5.94 kj/g FMI; mammalian carnivore (excluding fish eating), 16.8 kj/g DMI and 5.04 kj/g FMI; avian and reptilian carnivore (not fish), 15.4kJ/g DMI and 4.61 kj/g FMI; mammal eating a fish diet (piscivore), 18.7 kj/g DMI and 5.11 kj/g FMI; avian piscivore, 16.2 kj/g DMI and 4.43 kj/g FMI; herbivore (fermenter), 11.5 kj/g DMI and 3.80 kj/g FMI; herbivore (nonfermenter), 10.0 kj/g DMI and 3.30 kj/g FMI; omnivore, 14.0 kj/g DMI and 4.48 kj/g FMI; granivore, 16.9 kj/g DMI and 15.4 kj/g FMI (relatively high, due to the low water content of seeds); nectarivore, 16.0 kj/g DMI and 3.76 kj/g FMI; and frugivore, 6.6 kj/g DMI and 1.50 kj/g FMI. These factors were used to calculate all feeding rates reported in this review, even though more detailed conversion factors and feeding rate estimates are reported in a few of the research articles on individual species. The differences resulting from this simplification will have only a small influence on the regression of log-transformed data. The calculated feeding rates for 79 species of mammals, 95 species of birds, and 55 species of reptiles for which FMRs have been measured are shown in Table 1. Also shown are details regarding taxonomic affiliation (order or family), habitat, and diet.

6 5

7 6

8 7

9 8 These feeding rates are those needed to provide the metabolizable energy the animals burn in the field, so they represent steady-state conditions. Free-living animals that are growing or reproducing or storing fat for winter or migration will have feeding rates that are higher, perhaps even much higher, than estimated for the steady-state situation. Similarly, animals that are using body stores of energy during migration, rut, torpor, hibernation, etc. will have actual feeding rates that are lower than calculated, or even nonexistent. However, FMR data, and thus feeding rate estimates, from endothermic

10 9 animals undergoing starvation were not included in this analysis, nor were data from reptiles during inactive seasons (e.g. winter) or from juvenile birds or mammals that were not self-supporting. Regression analyses were done on the log10-transformed data for all mammals, all birds and all reptiles, as well as on every taxonomic, habitat and dietary category within those Classes, where sample sizes were adequate. Every regression for a category that was statistically significant, judging by a probability value < 0.05 according to an F-test for significance of the regression, is shown in Table 2 (mammals), Table 3 (birds), or Table 4 (reptiles). The regressions for desert marsupials (Equations 17 and 18) were calculated from new FMR data (Nagy and Bradshaw, 2000). Statistical tests to determine if allometric relationships in these tables differed from each other were beyond the scope of this review. Also not done were independent contrasts analyses (ICA), which adjust for phylogenetic relatedness (Garland et al., 1993). However, the FMR data on which the feeding rates in this article were based were subjected to independent contrasts analysis, and the reader is referred to Nagy et al. (1999) for details. In general, ICA yielded statistically similar slopes (a) and intercepts (b) for conventional FMR regressions, and the same result would be expected for ICA of the feeding rate regressions reported here. The probability values for the regressions for most groups in Tables 2-4 were quite low (<0.001), indicating that the relationships between log10 feeding rate and log10 body mass are robust for most groups. Some groups with smaller sample sizes (e.g. Pelecaniformes birds, with n = 4 and P = 0.031, Table 3) had much weaker relationships. The high coefficient of determination (r 2 ) values for many groups can be misleading. For example, the value for the All mammal DMI that indicates that variation in log10 body mass explains 94.7% of the variation in log10 DMI. In fact, variation in the untransformed data is much higher than this implies. The column in Tables 2-4 labeled Species deviation is the average absolute percent difference between the actual feeding rate for a species and the feeding rate calculated for that species (using the regression line value at its body mass; Speakman, 2000). If the DMIs for all 79 species of mammals were predicted from body mass values using Equation 1 (the All mammal group, Table 2) and compared to the

11 10 actual DMI values in Table 1, the average error (absolute error, ignoring sign) would be 41%. Scaling of feeding rate The allometric slope (b values) for feeding rate of the All mammal group (Table 2) is 0.74, substantially lower than the 1.0 value expected if there was a one-to-one relationship between food intake and body mass (i.e. a species ten times larger than another eats ten times as much food per day). To illustrate this, the scaling equation for DMI in All mammals, g DMI/d = 0.323(g body mass) 0.744, can be solved for two

12 11 representative mammals, one weighing 100 g and another weighing ten times more, or 1000g. The results are: predicted DMIs = 9.94 g/d and 55.1 g/d, respectively. The larger representative mammal should consume only 5.5 times more dry food daily (55.1/9.94 = 5.5), not ten times. The allometric slopes for many other mammalian, avian and reptilian groups are in or near the range of 0.6 to 0.9, but hummingbirds, desert lizards and lacertid lizards have slopes at or above 1.0 (Table 3 and 4), so a one-to-one relationship apparently does exist in these groups. Thus, with a few exceptions, we can say that among wild terrestrial vertebrates, larger species eat relatively less (kilogram for kilogram) than do their smaller relatives, while free-ranging in the field.

13 12 How do different groups compare? One way to facilitate comparing the different categories of vertebrates is first to account for body size differences by calculating expected DMI and FMI values for a common body mass. The Predicted food intake columns in Tables 2-4 show these values for a body mass of one kilogram in most cases, or in (parentheses) for either 50 g (some mammals and bird groups) or 10 g (some reptile groups) for those groups where typical body masses are low and one kilogram is outside the range of masses in that group. The common phrase to eat like a bird implies being very selective and eating only a small amount. In fact, Tables 2 and 3 reveal that a typical wild bird has a big appetite, consuming 31% more dry mass of food and 45% more fresh food each day than does a typical mammal (72 vs. 55 g DMI/d and 241 vs. 166 g FMI/d, respectively. The difference in food requirements between birds and reptiles is even more striking: a 1-kg

14 13 reptile ingests only 9% of the food, fresh or dry matter, each day as does a 1-kg bird. Similarly, a 1-kg mammal requires over eight times as much food per day to fuel its cost of living as does a 1-kg reptile, which may be living in the same habitat and eating a similar diet. Thus, among the terrestrial vertebrates, birds eat the most. In a similar way, we can compare the various groups of species within the Classes Mammalia, Aves and Reptilia. For example, in the Group deviation column of Table 2, the difference between the DMI rate predicted for a 1-kg eutherian mammal (60 g/d, from Eqn. 3) and the DMI rate for a 1-kg mammal from the All mammal group (55 g/d, Eqn. 1) is expressed as a percent of the All mammal prediction {100 [(DMIeuth DMIAll mam)/dmiall mam] = +9%). This method is not as good as comparing the predicted eutherian value with the 1-kg value calculated from the combined data for all noneutherian species, exclusive of the eutherian species, but such calculations for all the groups were beyond the scope of this review. Nevertheless, the Group deviation values can serve as relative indices. Among the mammal groups (Table 2), the Group deviation column suggests that eutherian mammals have somewhat higher feeding rates, and marsupials have somewhat lower feeding rates than all mammals combined, so that a 1-kg eutherian would have a daily feeding rate over 20% higher than a 1-kg marsupial. The seven species in the Order Carnivora have comparatively low field feeding rates. Desert mammals in general, and desert marsupials and desert rodents in particular apparently have relatively low food requirements for mammals. Similarly, insectivorous mammals and seed-eating (granivorous) mammals (many of whom are desert rodents) have relatively low daily food requirements. On the other hand, rodents in general, and especially mesic (moist habitat) rodents have relatively high feeding rates. Herbivorous mammals also have comparatively high food needs. Among birds, the Passerines (perching birds), the Apodidae (hummingbirds), the Pelecaniformes (gannets, tropicbirds), and especially the Charadriiformes (auks, gulls, shorebirds) have relatively high food requirements for birds (Table 3). Marine birds in general have somewhat elevated feeding rates, and temperate forest and desert birds apparently have reduced food needs. Regarding dietary categories, food requirements

15 14 seem comparatively low for omnivores, but rather high among carnivores and especially nectarivores. For the reptiles during their activity seasons (Table 4), several groups of lizards have somewhat elevated food requirements (all lizards, Scleroglossans, varanids and Lacertids), and herbivorous reptiles have comparatively low feeding rates. How to predict feeding rates To obtain an estimate of the daily food intake of a species of mammal, bird or reptile in its natural habitat, first check Table 1 to see if that species has been studied. If so, the estimates in Table 1 (or better, in the original research article describing that study, if included) will be the most reliable. If the species of interest has not been studied with doubly labeled water, its food requirements can be estimated by inserting its average body mass (in grams) into one or more of the allometric equations in Tables 2-4. For example, assume we wish to predict the fresh food intake for a common raven (Corvus corax, 866 g body mass), an omnivorous bird living in the Mojave Desert in California. The equation for All birds (Eqn. 36) becomes g FMI/d = ( 866) = = 218 g fresh matter intake per day. Equation 38 for Passerines, the taxon in which ravens belong, yields a prediction of 148 g FMI/d, Eqn. 56 for desert birds yields 104 g FMI/d, and the omnivorous bird equation (number 62) produces an estimate of 145 g FMI/d. Which estimate is the most reliable? The Passerine estimate is probably least accurate because the raven s body mass of 866 g is far outside the range of masses of species used to derive Equation 38 (6.6 to 96 g; Table 1), so a substantial extrapolation is involved, along with its attendant uncertainty. The desert bird equation is also suspicious because, although the raven in this example lives in a desert, common ravens are a widespread and often migratory species, so they may not show the reduced energy and food requirement possessed by desert specialist species, which contributed much data to the derivation of the equations for desert birds. This leaves the estimates of 218 (all birds) and 145 (omnivores), the latter being 33% lower than the former. The average error in the ability of the All birds equation to predict the feeding rates of the species used to derive the

16 15 equation is 40% ( Species deviation column, Table 3), and this error should be a conservative estimate of the error in predicting values for new species. The average error of prediction for the omnivorous bird equation is 33%. Thus, either estimate (218 or 145 g FMI/d) is within the range of error of the prediction from the other equation. Another way to evaluate the reliability of a prediction made from these equations is to calculate the 95% confidence intervals for the prediction. These values will be much larger than the average errors indicated in Tables 2-4. If desired, the 95% confidence intervals may be calculated from equations given in Nagy et al. (1999) for FMR predictions, and then converted to DMI or FMI equivalents using the appropriate conversion factors given above. The equations in Tables 2-4 yield estimated feeding rates that are needed for animals to obtain the metabolisable energy they used in their natural habitats, as determined with doubly labeled water. If the animal of interest to the reader is growing or reproducing or storing fat, its estimated feeding rate should be increased to include those avenues of metabolisable energy allocation. The literature on the species of interest or related species should be consulted to obtain rates of energy accumulation or allocation to production, which can then be added to estimated FMR. On the other hand, these equations will yield overestimates of food consumption for animals that are undergoing seasonal periods of weight loss due to relative or absolute starvation. Such periods include the nestling period for parent birds, the lactation period for nursing mammalian mothers, the migration period for many migratory terrestrial mammals, the cold seasons for temperate-zone reptiles, and the summer drought period for desert herbivores. Similarly, for wild animals held captive, such as in zoos, small outdoor enclosures or indoors in cages or pens, predicted feeding rates will probably be higher than actual food requirements. Probable reasons for this include: free-ranging animals must pay relatively higher costs of foraging for dispersed foods, avoiding or battling predators and parasites, dealing with more extreme climatic conditions, and interacting socially with conspecifics; and the foods given to captive animals are usually of higher quality (more metabolisable energy per gram of dry or fresh matter), so less biomass need be consumed, and this

17 16 reduces food requirements a second way, which is a reduced metabolic cost of food processing due to its greater digestibility. Conclusions Birds are the most expensive group of vertebrates on Earth. Kilogram for kilogram, a typical bird eats about 31% more food each day than does a mammal, and endotherms (birds and mammals) consume eight to eleven times as much food daily as does a reptile. Within these groups, feeding rates increase with increasing size of animal, but in a lessthan one-to-one manner, such that large animals use less food daily than that expected from their body mass (i.e. allometric slopes are usually less than 1.0). Feeding rates are strongly related to body mass within a variety of taxonomic, dietary and habitat groupings. The exponential (power) equations describing these relationships can be used to predict feeding rates in wild birds, reptiles, and mammals with an average error of about 40%, and an error as low as 5% in some groups. Such predictions should be adjusted up or down to account for higher expenses by breeding or growing animals or lower costs in captive animals. Acknowledgments I thank the University of California, Los Angeles for sabbatical leave time to write this review. I am grateful to Lisa Hazard and Danielle Shemanski for helpful comments on the manuscript. References Garland TJ, Dickerman AW, Janis CM, Jones JA (1993) Phylogenetic analysis of covariance by computer simulation. Systematic Biology 42, Lifson N, McClintock R (1966) Theory of use of the turnover rates of body water for measuring energy and material balance. Journal of Theoretical Biology 12,

18 17 Nagy KA (1983) The Doubly Labeled Water ( 3 HH 18 O) Method: A Guide to its Use. University of California Publication , USA, Los Angeles, University of California. Nagy KA, Bradshaw, SD (2000) Scaling of energy and water fluxes in free-living aridzone Australian marsupials. Journal of Mammalogy 81, (in proof) Nagy KA, Girard IA, Brown TK (1999) Energetics of free-ranging mammals, reptiles, and birds. Annual Review of Nutrition 19, Nagy KA, Peterson CC (1988) Scaling of Water Flux Rate in Animals. USA, Berkeley, University of California Press. Speakman JR (1997) Doubly Labelled Water: Theory and Practice. UK, London, Chapman and Hall. Speakman JR (2000) The cost of living: Field metabolic rates of small mammals. Advances in Ecological Research 30,